Wednesday, 19 March 2008

Linux on the desktop

I found Andrew Back's semi-rant about Linux on the desktop thought provoking. Back in 2003, in an idle moment, I contemplated starting a business based on helping small-to-medium sized businesses to deploy Linux on the desktop (hereinafter just "Linux"). And a friendly, highly competent, Linux-hugging sys admin guy I know (who shall remain nameless) said that "it's not ready for prime time". It stopped me dead in my tracks. Surely if he didn't think it was ready for prime time then I must be missing something?

There were a few key problems with Linux back then. Usability was an issue. Ease of switching was an issue. Choice was an issue.

So have the last five years made a difference?

Sure they have. Projects such as Ubuntu have a strong focus on usability now, and there have been big improvements as a result. Presumably the underlying code has improved. And yet there has been no great increase in usage.

In the same period, Apple is slowly gnawing into Microsoft's domination in the desktop market (hitting 6% in 2007 according to Wired, up from 3% in 2003). Why?

Also in the same period, an appetite for quality open source products such as Firefox and, to a lesser extent, VLC has developed, but not for Linux. Why?

Contrary to Mr.Back's views, I don't think the lack of device drivers is preventing people from switching to Linux. I don't hear many people saying that they'd switch if only they knew their devices would be supported. This is a red herring. It's something more fundamental than that. The problem is that most people still don't know or like Linux.

People (other than geeks) don't like learning new stuff, especially new computer programs if they can possibly help it. The exception to the rule: they will consider switching to an experience which is obviously more enjoyable or productive, and simple to grasp. Mac OS X is not only all three, but also this can be appreciated at a glance. It sells itself. People are keen to show their friends - both getting pleasure from the reaction. It's viral.

I'm not just talking about eye-candy here. The elements within Mac OS X that are pleasing on the eye have a primary purpose: productivity. Look at Exposé, or the Genie Effect, or dock magnification. They help you find what you're looking for quickly. The fact they look nice too is secondary.

Before this turns into a usability critique of Linux on the desktop, let's get back on track...

Linux has a long way to catch up with Apple and even Microsoft (Ow! Who threw that?) when it comes to overall usability. It would require a group of people to not only take responsibility for the usability of the OS (as the Ubuntu team have done so ably), but also responsibility for the over-arching usability, experience and inter-operability of all the apps that the target group would use. And I don't just mean bundling OpenOffice with your operating system. Case study: iLife. It isn't good enough that there are open source alternatives for iPhoto, or iTunes - if you want people to switch they've got to be better, and not just a little bit.

Of course, Apple's image has contributed to their success. But their recent success didn't run parallel to a change in image, which has remained pretty consistent throughout. That said, the image of Linux is bloody confusing to anyone taking an interest for the first time.

For Linux to break into the mainstream, I think it first needs to find a niche to carve out. And I think that niche is a particular type of business user. The type of business user who only needs an email client and a browser to get their job done. Ubuntu + Firefox (maybe using Google Docs) + Thunderbird (email client) is a solid suite, so start from there, consolidate, and then grow. The people using Linux in the workplace might - just might - make it their first choice at home too.

Let's accept that Linux isn't for everyone and stop calling it a battle. With patience and focus it will gradually become the OS of choice for a growing, vocal group of happy users (other than developers). And who knows where it might go from there?

Incidentally, I think OpenOffice should be treated as a prototype - throw it away and start again, and this time don't copy Microsoft Office 2000. Carve out a niche. Just my $0.02.

Disclaimer: I'm a Mac user and an open source fan. I'd love Linux to be good enough for me to switch to and am keen to support initiatives to improve Linux. I'd happily make the move to Linux if the combined hardware and software was better than Apple. Yes, really.

(Photo from MaxPa's flickrstream)

8 comments:

alex laurie said...

hi Phil, while i agree that things need to be a lot better there are just some fundamentals that a "business user" would want, typically that means proper MS Exchange (Lotus / Groupwise) integration for the Mail / PIM / Calendar client. (I know that IMAP sorts out email for Thunderbird).

I have found the only way i can achieve this on my various Laptops / Macs / Linux machines is via some form of mobile device acting as the conduit.

from a general ease of use pov I also agree that things have come on leaps and bounds with Linux, however as much as i would love to give my mum a Linux desktop (as all she needs is as per your post),I found myself replacing her PC with a Mac and have not had to teach her anything at all - it just works

BenJam said...

Curve-ball: Do you think the success of Microsoft and Apple based desktops can, in part, be attributed to the level of support provided by piers and, in the case of the maturing Apple, at the genius bar? Let's call it desktop herding...

Robbie said...

I must confess I'm a limbo user between XP and Ubuntu. I *love* Ubuntu and there's only a few small things that stop me moving over completely. I wrote about my experiences recently: http://blog.iclutton.com/2008/01/ubuntu-problem.html

I've never really used a Mac, I am always tempted, but then most of my apps run in a browser so it's just Firefox with different fonts right?

Andrew Back said...

Phil, some good insights but I think you are looking at this from a narrow perspective and one that is largely that of a power user who has come to expect the best in terms of usability. And for most people I would assert that Linux can meet with relative ease their needs in terms of usability.

With this in mind let us remember the wider problem domain: general purpose desktop computing. And feel I must ask you productivity speed-ups aside what requirement it is that Linux cannot meet? Sure, other operating systems are slicker but you pay for that and I'm not just talking 100 quid or so every time there is a major O/S update - more on this later. So I have yet to understand where Linux falls short here, unless you consider the advances in usability of the last few years that are present in the likes of OS X to be the mark of a capable general purpose computing platform. Most people just want basic tools on a solid platform that performs well. Linux can deliver this and much much more now.

This brings me on to the second point: license costs. We should not underestimate the cost of software licenses, E.g a case study via colleague Mark Kent: in a switch from Ofice to the still proprietary StarOffice Bristol City Council saved £1.1M over 5.5K desktops, and would save a further £186K if they moved to the open source oo.org that StarOffice is based upon.

People are coming to realise however that a reduction in software acquisition costs is a minor benefit compared to the removal of vendor lock-in. Here the benefit is two fold as you are able to not only break free from proprietary file formats, but are also no longer subjected to vendor driven mandatory hardware upgrade cycles. You get to work with open file formats and switch applications at will, and can choose to keep running on your 4+ year old hardware if you wish. Your Apple system may benefit from excellent usability but you are running on damned expensive and frankly overpriced hardware. Hardware that furthermore is often unreliable, E.g. the MacBook Pro and the ubiquitous click of death.

Granted Linux is confusing and work needs to be done on softer aspects such as branding/image and general accessibility. Ubuntu has made a good start here but more work needs to be done.

Contrary to Mr.Back's views, I don't think the lack of device drivers is preventing people from switching to Linux.

We'll have to agree to disagree and everyone's experiences will vary. But the biggest complaint I hear is from folks who installed Linux and a piece of hardware didn't work or took a lot of faff to get it to work. This can be infuriating and is enough to put off a lot of people for good.

For Linux to break into the mainstream, I think it first needs to find a niche to carve out. And I think that niche is a particular type of business user.

Computing geeks aside it has made big inroads in universities and is popular with artists and the third sector. It is also becoming fairly popular with the parents of geeks who have installed it back home so as to reduce support calls. Admittedly OS X would also suffer from less virii and trojans Etc., but Linux will happily run on old kit if required. Many parents approaching retirement age have little need for Core 2 Duo systems and would do better spending the cash on a holiday...

The type of business user who only needs an email client and a browser to get their job done.

Ouch. Linux can power: a general purpose desktop, a PDA, a smartphone, a supercomputer, a firewall, a router, a 10Gb/s switch, other major data centre and telco infrastructure, a video editing, audio or graphics workstation, and much more... And what it can't do currently it can be adapted to do as people have the tools and the freedom they need to adapt it.

@alexlaurie Have you tried Evolution for Exchange (inc. GAL via LDAP, and calendar) interoperability?

Phil Whitehouse said...

Andrew, thanks for your post. Some counter-comments:

"I think you are looking at this from a narrow perspective and one that is largely that of a power user"

I'm actually trying to look at this from a broad perspective; one of the average home Windows user who is terrified by the prospect of change. I have my reasons for not using Linux, but I didn't mention these in my post.

"I would assert that Linux can meet with relative ease their needs in terms of usability....most people just want basic tools on a solid platform that performs well."

I agree, but that's not really the issue. The issue is that people don't want to go through the pain of switching. I bought up Mac OS X because of the increase of market share, in stark comparison with Linux adoption. Most computer users have an OS that they've invested time in, are now comfortable with, and the alternatives have to be significantly better for change to even be contemplated. How else can one explain the contrast in adoption? As I said in my post, it ain't just image.

Licence costs, vendor lock in, open standards: average home users don't care. Businesses do, though - and that's why I think we should focus there.

Overall, I think it's dangerous to rely on the assumption that people ought to use Linux because we know what's good for them. We've got to get better at seeing things from the perspective of the user who hates change. Otherwise the stalemate will continue.

"People have the tools and the freedom they need to adapt it"

And for the large part, have no idea how to use these tools. And would probably be terrified by the fact that they even exist. It's important to note that what's fantastic for developers or system administrators is of zero interest or value to the common man.

I think Linux is great, and I acknowledge all the great things it can do. I'm asserting that, if greater Linux adoption is the goal, then the improved usability / improved device interoperability approach vector is the wrong one. Even reduced cost isn't enough to tempt most users, as Walmart has realised.

Andrew Back said...

Licence costs, vendor lock in, open standards: average home users don't care.

I beg to differ. Geeks tend to be fine with having to upgrade hardware and are less likely to baulk at software upgrade charges. Folks who just want their computer to work and remain supported are less likely to be happy with this. Hence many using pirated and often thus unprotected copies of Windows Etc. And as we become increasingly concerned with power consumption and waste efficient systems that consume less and are usable for longer periods will become more attractive.

Overall, I think it's dangerous to rely on the assumption that people ought to use Linux because we know what's good for them.

Granted. But Linux is becoming ever more compelling. And I'm sure there will always be proprietary alternatives even if it ticks most if not all the boxes with regards genuine requirements. As people buy into brands, fashion and in some cases needless desires. E.g. we don't need to drive at 160mph but there is a market for cars that will achieve such speeds.

And for the large part, have no idea how to use these tools. And would probably be terrified by the fact that they even exist.

Which is fine. I highlighted the freedom to innovate on Linux in response to your assertion that it is only good for basic tasks such as web browsing.

As an aside, that iPhone SDK is pretty good isn't it? The one that only lets you write crippled applications that won't threaten the business model of Apple and it's partners. Hrm.

Even reduced cost isn't enough to tempt most users, as Walmart has realised.

They still sell it online. But I have to concede that the loaded Linux distribution has been the subject of much criticism. There will be other budget Linux machines, the recipe will get refined and they will do better.

Jason Cunliffe said...

hmm... well Maybe the question is wrong. Seems like 'The Desktop' is the paradigm stuck truly still in the 80's.
[were about up to Amigas 1988 now]

OLPC
Hollwood Digital Filmmaking and renderfarms
Industrial systems and interfaces
TiVos
ADSL boxes
Sony Playstations...
Google
Gazillion web servers
Mobile cell thingy PDA.Phone.Cams.Personal media appliances

Linux is so huge and ubiquitous - visibly and invisibly.
The 'desktop' in a popular sense for many people has effectively already been replaced by web browsers runnign various apps [Google etc].

+A new announcement of a BitTorrent standalone whitebox implementation [running Linux] with the blessings of major film/tv networks.

Governments around the World are moving to Linux as their default and proscribed operating platform [Northern Europeans, China and others].

There have also been stagering improvements for digital artists of all ilk -- over paid and underpaid from Lord of the Rings budgets to VJs in hi-tech squats.

The Video/Audio/Midi FREE tools are now really viable which was not much true even 5 years ago. Those with no money can use pirated versions, but ther is no longer any excuse not to use Linux for fine art or commercial art.

The Homeschooling movement has been very strong in embracing Linux - but again, its more about content, web friendly, good openSource collabrative platforms and tools than it is about any particular OS.
Python, Ruby, Javascript, PHP coupled with APIs and importnatn apllications are what counts. The

OS or the desktop is less importnatn than ever before.
Multiswitching, VMWARE on linux etc. I read that many Micorsoft developers use VMWARE on linux as their foundation to test versions of their own OSs.

Adobe since buying Macromedia have been building up a storm around Flash, Flex, Javascript, AIR, MEdia Player etc. And keepin g good to their promise to support Linux. So far more effectively than Macromedia ever did.

Having evangelized all that - I'd argue strongly that what Linux needs is some very smart designers and artists to really think about Interface and usability afresh. REmember the good old days of CHI95 when there was so much ebnergy around interfaces and the near future. It all just got eaten up by browserwars and stupid wondws paradaigms being clones across all systems. Monoculture.

But since building architecture in the past few years has startyed to get really interesting, I'm hopefull Digital arhcitecture is gpoing to do the same. The right kind of tools and iunderstanding and Needs are noiw in place..

Elegant creative fun design approaches like TiddlyWiki, but scaled up that sort creative elegance to address freshly the user interface. Which can sit on top whatever then [Linux would be fine, maybe] - but hopefully we won't have to look at it or want to much.

...way back in the day Amigas had AreXX, a powerful interprocess scripting parsing language. Developers of paplciations just had to expose teh relvant parts of their API open some named I/O ports et voila, a thricing ecology of paplciatiosn talking to each other easily. New permutations asn uses emerged quickly sometimes beyond the imagiantions or intent of the individual designers. WE have that now pretty much with the web apis, especially since web2.0 techniques kicked in.

But I'd argue that what's been missing is the same sort of fluid interproceess communication to work across desktops, under them betwen them etc.
For example, PureData has nice way of doing that. TiddlyWiki does it too.
But it's the the desktop paradigm which is draggin us all down in a way. It creates OS wars when really it's not about that any more.

Another example, we about to see Javascript truly well embedded in servers. [like HELMA]
For a tool like TW that helps the scalability from private solo tool to public group tool. Anables roundtrip workflow development and permissions.

So how many functions does the OS need to handle which can not be dealt with in the same way?

Phil Whitehouse said...

"Seems like 'The Desktop' is the paradigm stuck truly still in the 80's."

Well, I'd agree there. But so long as there is a demand for a desktop OS, there will be a supply.